"Nash" Quiz mode is so desirable

[lumst]

There is an ICM Quiz software (Pokerstrategy ICM trainer), that calculated opponent ranges, using nash equilibrium.
However, it has a problem: it cannot quiz for custom payout/blind structures.
SNGWiz quiz can analyze any structure, but it doesnt calculate Nash equilibrium. It uses opponent models instead, it's similar to PT4.

Could PT4 Quiz implement both things: nash opponent ranges with custom structure ???

Below is explanation, why usual opponent models in SNGWiz and PT4 Quiz are wrong at some cases.
Let's investigate bubble stage for a little:
========
Payout structure 65/35, blinds 150/300/30, Hero is BU
1) stacks BU=1700, SB=700, BB=600 (notice: all effective stacks are 2BB)
ICM equities: 47.8%, 27.9%, 24.3%
Nash equilibrium is calculated http://www.holdemresources.net/hr/sngs/icmcalculator.html?action=calculate&bb=300&sb=150&ante=30&structure=0.65%2C0.35&s1=1700&s2=700&s3=600&s4=&s5=&s6=&s7=&s8=&s9=
PU___CA___OC____Range
BU_______________33.8%, 33+ Ax+ K3s+ K6o+ Q8s+ Q9o+ J9s+
_____SB__________5.4%, 99+ AJs+ AQo+
__________BB_____1.8%, JJ+
_____BB__________55.7%, 22+ Ax+ K2s+ K3o+ Q2s+ Q7o+ J3s+ J8o+ T5s+ T8o+ 95s+ 97o+ 85s+ 87o 74s+ 64s+ 53s+
SB_______________94.6%, 22+ 8x+ 72s+ 73o+ 62s+ 63o+ 52s+ 53o+ 42s+ 32s
_____BB__________100.0%, Any two

Now let's exchange stacks:
2) stacks BU=700, SB=1700, BB=600 (notice: all effective stacks are _still_ 2BB)
ICM equities: 27.9%, 47.8%, 24.3%
Nash equilibrium: http://www.holdemresources.net/hr/sngs/icmcalculator.html?action=calculate&bb=300&sb=150&ante=30&structure=0.65%2C0.35&s1=700&s2=1700&s3=600&s4=&s5=&s6=&s7=&s8=&s9=
PU___CA___OC____Range
BU_______________7.5%, 77+ ATs+ AJo+
_____SB__________4.2%, 99+ AQs+ AKo
__________BB_____0.9%, KK+
_____BB__________97.3%, 22+ 9x+ 82s+ 83o+ 72s+ 73o+ 62s+ 63o+ 3x+
SB_______________94.6%, 22+ 8x+ 72s+ 73o+ 62s+ 63o+ 52s+ 53o+ 42s+ 32s
_____BB__________96.4%, 22+ 8x+ 72s+ 73o+ 62s+ 63o+ 5x+ 42s+ 43o 32s
=========
If SB folds, we could see BU vs BB behaviors are very different in these examples. Could BB call range be approximated by current opponent model scheme, that has only inputs: number of players, position, effective stack?
I think: definitely NO.
Both examples has players:3, position: BB, effstack 2BB. But the result are different: 55.7% or 97.3%.

Why? Its simple. Opponent model should also be affected by ICM situation, and not only player count, position and effstack.

Let's investigate for a little: BU push, SB fold, BB think about calling. If BB wins BU,
in example 1)
the stacks will be 1100, 670, 1230.
ICM Equities: 36.4%,24.4%,39.1%.
BB needs odds against BU 24.3% : (39.1%-24.3%) = 1.64:1 (it's the bubble-factor value)
Thus, BB calls so tight, and BU know it, and could push looser

in example 2)
the stacks will be 100, 1670, 1230
ICM equities: 4.4%, 50.4%, 45.2%
BB needs odds against BU 24.3% : (45.2%-24.3%) = 1.16:1 (it's the bubble-factor value)
Thus, BB calls so loose, and BU know it, and must push tighter.

==========
So it's very desirable to have some nash mode of quiz - to learn how to play after Nash. Sure, in real play Hero should adjust his play against each real opponent style.

[kraada]

Thank you for this suggestion - I'll make certain the development team sees it.

[musaire]

Could PT4 Quiz implement both things: nash opponent ranges with custom structure ???

This would be too good to be true for me.

I know there is a way to import a hand to the offline ICM Nash Calculator beta software described here: http://www.holdemresources.net/hr/sngs/ ... lator.html

Features:

Offline Push/Fold Nash calculations for Windows, Mac OS X and Linux
Calculation of re-steal situations
Fully editable, unrestricted hand ranges
Hand import from HEM1 and PT4 databases
Additional equity models (PSM, Weitzman)
Automatic Updates

But it would be cool to launch a Nash analyzer right from the hands view in PT4 right-clicking a hand or an SnG .
At the moment for me to analyze an ICM situation I am looking at the tournament number in my PT3, then searching for it in SnG Wizard (very hard to do) and then sending a hand to the web based Nash calculator from there. Pretty laborious, isn't it... and I am doing it a lot.

If this is hard to implement it would be nice to have feature like it is in SitNGo Wizard - send a hand to the Nash calculator on the web, when viewing tournament/SnG hands in PT4.

It would be that good I'd hope any of my opponents wouldn't get hands on this software. Seriously.. :/


Just downloaded PT4 yesterday so I got no idea yet if there is any possibility to analyze an ICM situation (using opponent models) viewing tournament hands in PT4 right-clicking a hand or tournament at all or is there a calculator only with no option to import a hand from database...

[musaire]

I guess I wasn't talking about the quiz. I was talking about the possibilities to analyze the hands from the database.
Implementing it in quiz is not bad either though but we'd still need a software to analyze the played hands as well.

[kraada]

This kind of thing is on our own future wishlist, but at the moment we are working to get everything stable for a commercial release, so I can't say at the moment when we'll be looking at this kind of thing.

[pt4pt4pt4]

I would suggest a ICM Quiz that is devolped from the users biggest mistakes.

PT4 could go through your database (filteres by date) and review your biggest mistakes and let you try them again.

[kraada]

Our ICM quiz is currently configurable - you can pick the structure, pick your position, and the range of cards you want to have in the quiz, as well as set a minimum and maximum number of players.

So if you want to only have a quiz where you have AQo you can do that if you want, or if you only want bubble hands you can do that too.

I agree something like "give me a quiz in the situations similar to my biggest mistakes" would be cool, but it would take a lot of analysis to find and track those spots, so it would be a pretty big addition.

[SeBiX7]

This kind of thing is on our own future wishlist, but at the moment we are working to get everything stable for a commercial release, so I can't say at the moment when we'll be looking at this kind of thing.

bump :)